Geospatial Data Science course materials, taught once and left open

Fourteen units, from geometry to mobility

The repository is organised as a course, one directory per unit, and the topic list in the documentation is a decent syllabus on its own. The sequence runs: geometric objects, geospatial data in Python, choropleth mapping, spatial weights, spatial autocorrelation, spatial clustering, point pattern analysis, OpenStreetMap and OSMnx, spatial networks, bicycle networks, individual mobility, mobility patterns, aggregate mobility and urban scaling, and finally sustainable mobility and geospatial epidemiology. On disk each unit is its own numbered directory, alongside a `docs/` folder for supporting material and an `admin/` directory. The unit numbering matters for anyone adapting this: it is a taught progression, where spatial weights lead into spatial autocorrelation and then clustering, and where individual mobility patterns are needed before aggregate mobility can be discussed. The materials were slightly improved and reordered after the course ran, so the on-disk state is a cleaned-up version of what the students saw.

The prerequisites decide who can follow it

The documentation states the prerequisites as basics in data science, including statistics, Python and pandas, and puts the ideal level at a first year Master in Data Science. That is a useful filter, because the course is a bridge from general data science into spatial methods rather than an introduction to Python. A course page is published on the IT University of Copenhagen learning platform for the original run. There is also a schedule, published as an image in the documentation folder, which tells you how the lectures were spread across the term. For a self-learner the practical consequence is that the ordering matters and the assumed statistical background does too: the spatial autocorrelation and clustering units build on concepts the reader is expected to already have. The repository also carries a `CITATION.cff` file, which is a small sign that the material is meant to be cited when reused in teaching or writing.

The sources section says which book each unit leans on

One of the most useful pages in the repository is the sources list, which attributes the material and gives approximate percentages for the main sources. Arribas-Bel's Geographic Data Science accounts for 40 percent, and Alessandretti and Szell's Urban Mobility for 25 percent, with the latter developed after the course, which tells you that the later mobility units are more advanced than the earlier ones. Tenkanen's Automating GIS-processes and Boeing's OSMnx notebooks each account for 7 percent, Tenkanen's pyrosm 1 percent, and the spaghetti and scikit-mobility tutorials 2 percent each. A longer list of other major sources names the Geographic Data Science with Python book, the PySAL spaghetti library, scikit-mobility tutorials, the transport geography reference, the spatial networks book, the human mobility review, Batty on cities, and a long run of individual papers on mobility patterns. The stated framing is that the materials stand on the shoulders of giants, and more sources are referenced inside the slides and notebooks themselves, so the bibliography is a research reading list as much as an attribution list.

The exercises and tutorials are a separate repository

If you plan to work through this course, the exercises are not in the same place. The documentation points to a separate repository, GDS2022_exercises, for the exercise materials and tutorials, and credits them to Ane Rahbek Vierø and Anastassia Vybornova, while the lectures themselves are credited to Michael Szell. That split is a practical detail worth planning around, because a course without its exercises is half a course, and the two pieces can drift apart. The acknowledgements also mention help with the syllabus, the exam form and the project description from Adéla Sobotkova, and slide materials from Vedran Sekara, which tells you the course included a project and an exam rather than being only notebooks. A researcher looking for reusable code should also note that the licences of the underlying libraries and of the course material itself are separate matters, and the documentation says to consult the licence of the respective source for other reuse questions.

The offshoots show what the course turned into

The documentation lists what grew out of the material after it was taught, under the heading Offshoots. There is a published paper on teaching spatial data science by Vierø and Szell, which is the closest thing to a methodological write-up of the course design itself. There is the Urban Mobility preprint on arXiv, developed after the course. There is a GDS 2023 branch in the same repository, which suggests the material has been revised for a later run. And there is a separate Geospatial Data Science Tutorial repository described as the course condensed to three hours, which is the most interesting of the four for anyone with a short workshop slot. Taken together, these entries tell you that this is a living body of teaching material with a publication record rather than a one-off upload of lecture slides, and that the author has been willing to package it in different shapes for different audiences.

Reuse terms are narrower than the licence badge

The licence situation needs reading carefully, because two different signals appear. The project listing shows CC-BY-4.0 as the licence, while the licence section in the documentation says something more restrictive: that all materials were used for educational, non-commercial reasons only, that you are free to use them as you wish for the same purpose at your own risk, and that for other reuse questions you should consult the licence of the respective source. It then explains the choice: the main sources use the CC BY-SA 4.0 licence, so this material uses it too. The practical reading is that reuse in a commercial setting, or in a product, is not what these terms grant, and that the upstream material has its own conditions. This is not legal advice, but it is the difference between a badge you can act on and a paragraph you have to read, and any organisation planning to fold these notebooks into a commercial pipeline needs to resolve it with the author first.

A 2022 course with a 2026 push and no releases

The maintenance picture is straightforward and unusual. The course was held for the first time in spring 2022, the repository has no GitHub releases, and the last push to the repository was on 2026-04-20 with the repository not archived. So the material is a snapshot of a 2022 course that has been touched since, including the 2023 branch and the three hour tutorial, rather than a curriculum under annual revision. For a learner, that is mostly fine, because the core methods in the earlier units, spatial weights, autocorrelation, point pattern analysis, are stable subjects and the teaching is in the notebooks rather than in a moving API. For anyone planning to teach from it, the practical risk is the toolchain, since the notebooks lean on OSMnx, PySAL and related libraries whose interfaces have moved since 2022, and the repository has no release tags to tell you which combination was known to work. The topic tags reflect the same balance, listing course materials, geospatial analysis, street networks and OpenStreetMap rather than anything about an active software project.

Editorial conclusion

Learners who want a structured path from geometric objects to urban mobility, and who are comfortable with statistics, Python and pandas, will find these notebooks a rare free curriculum; the sources section even tells you which external book each unit leans on. It is the wrong starting point if you need a single technique, since the material is organised as a course rather than a reference, and the exercises live in a separate repository you have to fetch. Before you build on it, read the licence paragraph carefully, since the stated reuse terms are for educational and non-commercial purposes, and check the `unit08_openstreetmap/` and `unit09_spatialnetworks/` notebooks against current OSMnx releases, as the course dates from 2022.

Frequently asked questions

Official sources

1. Issues
2. License: CC-BY-4.0
3. mszell/geospatialdatascience on GitHub
4. README